Morphosyntax of Modals and Quasimodals(1) - DSpace at Waseda ...
Morphosyntax of Modals and Quasimodals(1) - DSpace at Waseda ...
Morphosyntax of Modals and Quasimodals(1) - DSpace at Waseda ...
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
<strong>Morphosyntax</strong> <strong>of</strong> <strong>Modals</strong> <strong>and</strong><br />
<strong>Quasimodals</strong> <br />
Introduction<br />
Ryo OTOGURO<br />
Modality is a gramm<strong>at</strong>ical c<strong>at</strong>egory where language’s morphology,<br />
syntax, semantics <strong>and</strong> pragm<strong>at</strong>ics meet. Due to its ubiquitous n<strong>at</strong>ure,<br />
linguistic phenomena covered by the term ‘modality’ are diverse.<br />
Traditionally modality is <strong>of</strong>ten defined as the expression <strong>of</strong> the<br />
speaker’s <strong>at</strong>titude, opinions or emotions towards the proposition, or<br />
more generally the subjective expression e.g. Lyons , Bybee et al.<br />
. For instance, a pair <strong>of</strong> sentences in are concerned with the<br />
speaker’s judgement <strong>of</strong> the proposition th<strong>at</strong> K<strong>at</strong>e is <strong>at</strong> home. The pair<br />
in , on the other h<strong>and</strong>, are relevant to the speaker’s <strong>at</strong>titude<br />
towards a potential future event, namely K<strong>at</strong>e’s coming in. The former<br />
is called epistemic modality, whereas the l<strong>at</strong>ter is called deontic<br />
modality Palmer : .<br />
a. K<strong>at</strong>e may be home now.<br />
It is possible possibly the case th<strong>at</strong> K<strong>at</strong>e is <strong>at</strong> home<br />
now.<br />
<strong>Morphosyntax</strong> <strong>of</strong> <strong>Modals</strong> <strong>and</strong> <strong>Quasimodals</strong> +.*<br />
b. K<strong>at</strong>e must be home now.<br />
It is necessarily the case th<strong>at</strong> K<strong>at</strong>e is <strong>at</strong> home now.
+-3<br />
a. K<strong>at</strong>e may come in now.<br />
It is possible for K<strong>at</strong>e to come in now.<br />
b. K<strong>at</strong>e must come in now.<br />
It is necessary for K<strong>at</strong>e to come in now.<br />
Although epistemic <strong>and</strong> deontic modality is encoded explicitly by<br />
modal auxiliaries in <strong>and</strong> , the definition based on<br />
subjectivity do not stop us from including a wider variety <strong>of</strong> syntactic<br />
constructions as modal expressions than expected. As Narrog <br />
points out, passive voice <strong>and</strong> certain aspectual expressions may<br />
convey the speaker’s <strong>at</strong>titude towards the propositions Kroeger :<br />
, Narrog :, . <br />
<strong>Morphosyntax</strong> <strong>of</strong> <strong>Modals</strong> <strong>and</strong> <strong>Quasimodals</strong><br />
a. Korede amenidemo huraretara,<br />
thison rainDATEXPL rainPASSCOND<br />
sanz<strong>and</strong>esuyo.<br />
aufulCOP.NPSTEXCL<br />
‘If we got rained on, th<strong>at</strong> would be awful.’<br />
b. Tomo kem<strong>at</strong>ian anak.<br />
Tomo die.ADV child<br />
‘Tomo suffered the de<strong>at</strong>h <strong>of</strong> a child.’<br />
You’re telling me you don’t love me anymore.<br />
a <strong>and</strong> b are advers<strong>at</strong>ive passives in Japanese <strong>and</strong> Malay<br />
respectively. Unlike ordinary passive form<strong>at</strong>ion, which involves
decrease <strong>of</strong> the valency <strong>of</strong> the predic<strong>at</strong>e, an advers<strong>at</strong>ive passive<br />
pragm<strong>at</strong>ically adds the speaker’s neg<strong>at</strong>ive feeling to the literal<br />
meaning <strong>of</strong> the proposition. is wh<strong>at</strong> Wright calls<br />
experiential progressives. The progressive aspect may convey the<br />
speaker’s epistemic stance with regard to the interlocutor’s utterance.<br />
In , the speaker neg<strong>at</strong>ively evalu<strong>at</strong>es the interlocutor’s<br />
utterance <strong>of</strong> not loving the speaker anymore. Those constructions<br />
clearly deliver the speaker’s subjective evalu<strong>at</strong>ion <strong>of</strong> the proposition,<br />
<strong>and</strong> according to the definition <strong>of</strong> modality given above, they may be<br />
c<strong>at</strong>egorised as modal expressions.<br />
The fact th<strong>at</strong> a language has various means to express the speaker’s<br />
subjectivity, however, does not mean all <strong>of</strong> them are directly encoded<br />
in the language’s gramm<strong>at</strong>ical system. As argued in a number <strong>of</strong><br />
studies, linguistically more salient definition <strong>of</strong> modality is <strong>of</strong>fered<br />
based on ‘factuality’, namely realis/irrealis distinction e.g. Lyons<br />
, Kiefer , Mithun , Palmer , Narrog . One<strong>of</strong> the<br />
crucial aspects <strong>of</strong> factualitybased definition <strong>of</strong> modality is its<br />
independence from the speaker’s point <strong>of</strong> view since factuality is a<br />
semanticoriented notion. Narrog :, for instance, proposes<br />
the following definition :<br />
<strong>Morphosyntax</strong> <strong>of</strong> <strong>Modals</strong> <strong>and</strong> <strong>Quasimodals</strong> +-2<br />
Modality is a linguistic c<strong>at</strong>egory referring to the factual<br />
st<strong>at</strong>us <strong>of</strong> a st<strong>at</strong>e <strong>of</strong> affairs. The expression <strong>of</strong> a st<strong>at</strong>e <strong>of</strong> affairs<br />
is modalised if it is marked for being undetermined with<br />
respect to its factual st<strong>at</strong>us, i.e. is neither positively nor<br />
neg<strong>at</strong>ively factual.
+-1<br />
This definition properly excludes the cases like <strong>and</strong> as both<br />
advers<strong>at</strong>ive passives <strong>and</strong> experiential progressives are clearly factual,<br />
which is outside the domain <strong>of</strong> modality.<br />
The exponents encoding indeterminacy <strong>of</strong> factuality still varies<br />
across languages. In English examples <strong>and</strong> , for example, the<br />
exponents are modal auxiliaries. Germanic languages dominantly use<br />
auxiliary verbs to express modality, such as English can, German<br />
können, Norwegian kan <strong>and</strong> Frisian kin, all <strong>of</strong> which are historically<br />
derived from Gothic kunnan ‘know’ Harbert . Another p<strong>at</strong>tern is<br />
verb inflection, <strong>of</strong>ten called mood, as in indic<strong>at</strong>ive <strong>and</strong> subjunctive<br />
forms in Spanish <strong>and</strong> an irrealis form in Amele <br />
Palmer : .<br />
<strong>Morphosyntax</strong> <strong>of</strong> <strong>Modals</strong> <strong>and</strong> <strong>Quasimodals</strong><br />
a. Creo que aprende.<br />
I believe th<strong>at</strong> learn..SG.PRES.IND<br />
‘I believe th<strong>at</strong> he is learning.’<br />
b. Dudo que aprenda.<br />
I doubt th<strong>at</strong> learn..SG.PRES.SUBJUNCT<br />
‘I doubt th<strong>at</strong> he’s learning.’<br />
Ho bubusaleb age qoqagan.<br />
pig SIMrun out.SG.DS.IRR .PL hit.PLFUT<br />
‘They will kill the pig as it runs out.’<br />
Although the surface descriptive fact is rel<strong>at</strong>ively straightforward,<br />
th<strong>at</strong> is auxiliaries are syntactically independent lexical items th<strong>at</strong> take
<strong>Morphosyntax</strong> <strong>of</strong> <strong>Modals</strong> <strong>and</strong> <strong>Quasimodals</strong> +-0<br />
a verbal complement whereas verb inflectional form<strong>at</strong>ives are<br />
realis<strong>at</strong>ions <strong>of</strong> morphological c<strong>at</strong>egories constructing paradigm<strong>at</strong>ic<br />
organis<strong>at</strong>ion <strong>of</strong> a given lexeme, it is challenging for any gramm<strong>at</strong>ical<br />
theories to capture the morphosyntactic n<strong>at</strong>ure <strong>of</strong> such a diverse<br />
range <strong>of</strong> encoding p<strong>at</strong>terns. With regard to auxiliaries, they are<br />
obviously syntactically independent units, but whether they are<br />
genuine argumenttaking predic<strong>at</strong>es or not requires a careful<br />
inspection <strong>of</strong> the empirical d<strong>at</strong>a. Another issue is phrase structure<br />
wordhood <strong>of</strong> the form<strong>at</strong>ives. Although some verbal inflectional forms<br />
are clearly affixes as illustr<strong>at</strong>ed in Spanish <strong>and</strong> Amele examples, there<br />
are marginal cases in which the boundary between auxiliarylike<br />
elements <strong>and</strong> affix/cliticlike elements are not straightforwardly<br />
identifiable. The primary aim <strong>of</strong> the present study is to examine<br />
whether a unified framework for modality can be provided in<br />
constraintbased parallel gramm<strong>at</strong>ical architecture, Lexical Functional<br />
Grammar Bresnan . To this end, I will briefly overview the<br />
historical background <strong>of</strong> the analyses th<strong>at</strong> have been proposed in the<br />
gener<strong>at</strong>ive liter<strong>at</strong>ure in section . Ithen present the insightful d<strong>at</strong>a in<br />
Japanese, focusing on the c<strong>at</strong>egorical differences between socalled<br />
genuine modals <strong>and</strong> quasimodals in section . In section , I shall<br />
summarise the previous LFG account for English modals, <strong>and</strong> then<br />
illustr<strong>at</strong>e how LFG gives us a unified framework for Japanese d<strong>at</strong>a as<br />
well. The discussion will be concluded in section .<br />
Fe<strong>at</strong>ure or predic<strong>at</strong>e?<br />
<strong>Modals</strong> along with tense <strong>and</strong> aspect have <strong>at</strong>tracted much <strong>at</strong>tention<br />
since early days <strong>of</strong> gener<strong>at</strong>ive syntax. As summarised in Falk ,
+-/<br />
the analyses are mainly divided into two types : AUX fe<strong>at</strong>ure analysis<br />
<strong>and</strong> AUX predic<strong>at</strong>e analysis. The AUX fe<strong>at</strong>ure analysis is originally<br />
proposed by Chomsky <strong>and</strong> its fundamental assumption is th<strong>at</strong><br />
an auxiliary is an element th<strong>at</strong> merely contributes tense, aspect <strong>and</strong><br />
mood fe<strong>at</strong>ures to the predic<strong>at</strong>eargument rel<strong>at</strong>ionships specified by<br />
the lexical verb. Therefore, a cluster <strong>of</strong> auxiliaries are loc<strong>at</strong>ed under a<br />
single node called AUX as in .<br />
<br />
<strong>Morphosyntax</strong> <strong>of</strong> <strong>Modals</strong> <strong>and</strong> <strong>Quasimodals</strong><br />
is a phrase structure tree for a sentence, the author may have<br />
written th<strong>at</strong> novel. The tense, modal <strong>and</strong> aspectual auxiliaries alongside<br />
averbal suffix are loc<strong>at</strong>ed all under the AUX node.<br />
According to the development <strong>of</strong> transform<strong>at</strong>ional frameworks, the<br />
AUX fe<strong>at</strong>ure analysis has been revised, in which an AUX element is<br />
regarded as a head <strong>of</strong> the entire clause, namely the Inflection heads<br />
the IP projection S. IP is called functional projection <strong>and</strong> l<strong>at</strong>er<br />
further divided into a layer <strong>of</strong> functional projections, AgrS, AgrO <strong>and</strong><br />
T Pollock . Some recent version <strong>of</strong> transform<strong>at</strong>ional grammar,<br />
called Cartography <strong>of</strong> grammar, proposes even more fine grained<br />
functional projections. Cinque , for example, assumes a<br />
structure in which various types <strong>of</strong> Modal, Tense <strong>and</strong> Aspect
are organised in a hierarchical way ; a simplified version can be<br />
represented as in the subscripts represent different types <strong>of</strong><br />
functional projections.<br />
<br />
Despite the different structural organis<strong>at</strong>ions, namely a single AUX<br />
node <strong>and</strong> a layer <strong>of</strong> functional projections, the essential part <strong>of</strong> the<br />
AUX fe<strong>at</strong>ure analysis is the claim th<strong>at</strong> fe<strong>at</strong>ures like tense, aspect, <strong>and</strong><br />
modality are all functional, i.e. nonpredic<strong>at</strong>ive, elements in the<br />
syntactic structure.<br />
The AUX predic<strong>at</strong>e analysis, on the other h<strong>and</strong>, assumes th<strong>at</strong> an<br />
auxiliary is one type <strong>of</strong> predic<strong>at</strong>es <strong>and</strong> takes a complement as in <br />
cf. Ross <br />
<strong>Morphosyntax</strong> <strong>of</strong> <strong>Modals</strong> <strong>and</strong> <strong>Quasimodals</strong> +-.
+--<br />
<br />
<strong>Morphosyntax</strong> <strong>of</strong> <strong>Modals</strong> <strong>and</strong> <strong>Quasimodals</strong><br />
In contrast to , the two auxiliaries both take a sentential<br />
complement. The subject, the author, origin<strong>at</strong>es in a lower subject<br />
position <strong>of</strong> the lexical verb, written, moves to the subject position <strong>of</strong><br />
the lower auxiliary, have, <strong>and</strong> further moves up to the subject position<br />
<strong>of</strong> the higher auxiliary, may. Hence, the AUX predic<strong>at</strong>e analysis<br />
regards auxiliaries a kind <strong>of</strong> raising predic<strong>at</strong>es.<br />
Although the recent transform<strong>at</strong>ional grammar dominantly follows<br />
the AUX fe<strong>at</strong>ure analysis as most dram<strong>at</strong>ically exemplified in , in<br />
principle the choice between the two competing analyses is<br />
determined based on conceptual arguments <strong>and</strong> empirical observ<strong>at</strong>ion,<br />
<strong>and</strong> it <strong>of</strong>ten varies across languages <strong>and</strong> even within a single<br />
language. For instance, Falk argues th<strong>at</strong> an English copul<strong>at</strong>ive<br />
progressive auxiliary must be regarded as a predic<strong>at</strong>e, citing<br />
Jackend<strong>of</strong>f’s , illustr<strong>at</strong>ion <strong>of</strong> paradigm<strong>at</strong>ic contrast <strong>of</strong> a<br />
progressive auxiliary to other lexical verbs as in <strong>and</strong> the<br />
availability <strong>of</strong> a PP complement both for the lexical verbs <strong>and</strong> the<br />
progressive auxiliary in a parallel fashion as in .
a. The children were e<strong>at</strong>ing breakfast.<br />
being in the st<strong>at</strong>e <strong>of</strong> e<strong>at</strong>ing breakfast<br />
b. The children started e<strong>at</strong>ing breakfast.<br />
entering the st<strong>at</strong>e <strong>of</strong> e<strong>at</strong>ing breakfast<br />
c. The children kept e<strong>at</strong>ing breakfast.<br />
continuing the st<strong>at</strong>e <strong>of</strong> e<strong>at</strong>ing breakfast<br />
d. The children stopped e<strong>at</strong>ing breakfast.<br />
leaving the st<strong>at</strong>e <strong>of</strong> e<strong>at</strong>ing breakfast<br />
a. John kept Bill running/<strong>at</strong> a run.<br />
b. Moe went on working/with his work.<br />
c. Rodgers is working/<strong>at</strong> work on a new play.<br />
The empirical d<strong>at</strong>a in <strong>and</strong> seem to suggest th<strong>at</strong> the English<br />
progressive auxiliary is able to take a complement <strong>and</strong> this fact<br />
cannot be captured by the AUX fe<strong>at</strong>ure analysis.<br />
Norwegian provides even more robust evidence for the predic<strong>at</strong>ive<br />
n<strong>at</strong>ure <strong>of</strong> auxiliaries. As illustr<strong>at</strong>ed in , apronominal can appear<br />
as a complement <strong>of</strong> a modal <strong>and</strong> tense auxiliary in Norwegian. The<br />
d<strong>at</strong>a strongly suggest th<strong>at</strong> modal <strong>and</strong> tense auxiliaries in Norwegian<br />
are fullfledged argumenttaking predic<strong>at</strong>es. Dyvik , cited in<br />
Falk .<br />
<strong>Morphosyntax</strong> <strong>of</strong> <strong>Modals</strong> <strong>and</strong> <strong>Quasimodals</strong> +-,<br />
a. Jeg vi/kan/ma˚/skal dette.<br />
I will/can/may/shall this<br />
‘I want/am able to do/am obliged to do/have a duty to do
+-+<br />
this.’<br />
b. Vil det regne? Det vil det.<br />
will it rain? it will th<strong>at</strong><br />
‘Will it raining? It will th<strong>at</strong>.’<br />
Romance aspectual auxiliaries, on the contrary, do not behave like<br />
predic<strong>at</strong>es. For instance, although Spanish allows an infinitival<br />
complement to be fronted leaving the main verb behind as in a,<br />
the same oper<strong>at</strong>ion is not available for the complement <strong>of</strong> a perfective<br />
auxiliary as illustr<strong>at</strong>ed by the ungramm<strong>at</strong>icality <strong>of</strong> b Schwarze<br />
, cited in Falk .<br />
<strong>Morphosyntax</strong> <strong>of</strong> <strong>Modals</strong> <strong>and</strong> <strong>Quasimodals</strong><br />
a. Ver el castillo, lo quiere.<br />
to see the castle it.ACC want..SG<br />
‘To see the castle, he wants it.’<br />
b. Visto el castillo, lo ha.<br />
seen the castle it.ACC AUX..SG<br />
‘Seen the castle, he has it.’<br />
The empirical observ<strong>at</strong>ion reveals th<strong>at</strong> whether auxiliary elements<br />
should be analysed as genuine argumenttaking predic<strong>at</strong>es or mere<br />
fe<strong>at</strong>ure contributors is largely an itemspecific m<strong>at</strong>ter. This suggests<br />
th<strong>at</strong> gramm<strong>at</strong>ical theory must be able to provide a framework th<strong>at</strong><br />
accommod<strong>at</strong>es both types <strong>of</strong> analyses in its design.<br />
<strong>Modals</strong> <strong>and</strong> quasimodals<br />
As pointed out in section , the distinction between auxiliaries <strong>and</strong>
verbal affixes is not always clearcut. Japanese is one <strong>of</strong> the<br />
languages th<strong>at</strong> develop a diverse range <strong>of</strong> modal expressions, <strong>and</strong> on<br />
the surface, they look all similar Nitta : , Ueda : <br />
.<br />
a. amewa furumai.<br />
rainTOPIC fallNAG.CONJECT<br />
‘It may not rain.’<br />
b. asuwa harerudesyoo.<br />
tomorrowTOPIC sunnyCONJECT.POL<br />
‘Probably, it will be sunny tomorrow.’<br />
c. hayaku tabero/tabetamae/tabenasai.<br />
quick e<strong>at</strong>IMP<br />
‘E<strong>at</strong> quickly.’<br />
<strong>Morphosyntax</strong> <strong>of</strong> <strong>Modals</strong> <strong>and</strong> <strong>Quasimodals</strong> +-*<br />
d. w<strong>at</strong>asiga kono hono yakusoo.<br />
INOM this bookACC transl<strong>at</strong>eINTENT<br />
‘I am willing to transl<strong>at</strong>e this book.’<br />
a. Tarooga daigakue singaku sisooda.<br />
TaroNOM universityto enter doREPORT<br />
‘It seems to be the case th<strong>at</strong> Taro will go to the university.’<br />
b. Taroowa daigakue singaku subekida.<br />
TaroTOPIC universityto enter doOBLIG<br />
‘Taro should go to the university.’<br />
c. sy<strong>at</strong>yooga kuruyooda.<br />
CEONOM comeCONJECT<br />
‘The CEO will probably come.’
+,3<br />
The italicised exponents all encode modality <strong>and</strong> follow a lexical<br />
verb. However, as Nitta points out, the modal elements in <br />
<strong>and</strong> are called ‘genuine modals’ <strong>and</strong> ‘quasimodals’<br />
respectively ,<strong>and</strong> they exhibit different morphosyntactic behaviours<br />
as summarised in table .<br />
Modal Quasimodal<br />
Tense distinction No Yes<br />
Polarity distinction No Yes<br />
Stacking No Yes<br />
Firstly, quasimodals can either be past or nonpast tense forms while<br />
genuine modals cannot. illustr<strong>at</strong>es th<strong>at</strong> quasimodals have tense<br />
distinctions, sooda/sood<strong>at</strong>ta <strong>and</strong> bekida/bekid<strong>at</strong>ta.<br />
a. Taroowa daigakue singakusisooda/<br />
TaroTOPIC universityto enter doREPORT.NONPAST/<br />
sood<strong>at</strong>ta.<br />
REPORT.PAST<br />
‘It seems/seemed to be the case th<strong>at</strong> Taro will/would go to<br />
the university.’<br />
b. Taroowa daigakue singaku subekida/<br />
TaroTOPIC universityto enter doOBLIG.NONPAST/<br />
bekid<strong>at</strong>ta.<br />
OBLIG.PAST<br />
<strong>Morphosyntax</strong> <strong>of</strong> <strong>Modals</strong> <strong>and</strong> <strong>Quasimodals</strong><br />
Table . Modal <strong>and</strong> quasimodal<br />
‘Taro should go/should have gone to the university.’
Secondly, quasimodals can be neg<strong>at</strong>ed as shown in , whereas<br />
the neg<strong>at</strong>ive polarity can only be expressed by inherently neg<strong>at</strong>ive<br />
form<strong>at</strong>ives in genuine modals cf. a.<br />
a. Taroowa daigakue singaku sisoodenai.<br />
TaroTOPIC universityto enter doREPORT.NEG<br />
‘It doesn’t seem to be the case th<strong>at</strong> Taro will go to the<br />
university.’<br />
b. Taroowa daigakue singaku subekidenai.<br />
TaroTOPIC universityto enter doOBLIG.NEG<br />
‘Taro should not go to the university.’<br />
Finally, quasimodals can appear multiple times as long as the<br />
semantic combin<strong>at</strong>ions are interpretable as in a. However, such<br />
stacking <strong>of</strong> modals is impossible for genuine modals as shown in .<br />
a. sy<strong>at</strong>yooga koozyooni kurutumorid<strong>at</strong>tarasii.<br />
CEONOM factoryto comeINTENT.PASTREPORT<br />
‘I heard th<strong>at</strong> the CEO was willing to come to the factory.’<br />
b. kono yamani noboroomai.<br />
this mountainLOC climbINTENTNEG.CONJUNCT<br />
‘It may not be the case th<strong>at</strong> we are willing to climb this<br />
mountain.’<br />
<strong>Morphosyntax</strong> <strong>of</strong> <strong>Modals</strong> <strong>and</strong> <strong>Quasimodals</strong> +,2<br />
All those morphosyntactic behaviours indic<strong>at</strong>e th<strong>at</strong> quasimodals<br />
are syntactically independent predic<strong>at</strong>ive verbs in th<strong>at</strong> they inflect for<br />
tense <strong>and</strong> polarity in the same way as other ordinary lexical verbs. In
+,1<br />
addition, the stacking fact strongly supports the idea th<strong>at</strong> they take a<br />
syntactic complement as assumed in the AUX predic<strong>at</strong>e analysis.<br />
The genuine modals, on the other h<strong>and</strong>, exhibit the properties <strong>of</strong><br />
morphological products. The past/nonpast tense distinction<br />
expresses declar<strong>at</strong>ive mood, which is in the same morphosyntactic<br />
c<strong>at</strong>egory, MOOD, asother genuine modals. It also explains why genuine<br />
modals cannot cooccur with each other. Since they particip<strong>at</strong>e in the<br />
same inflectional c<strong>at</strong>egory, the cooccurrence <strong>of</strong> more than one<br />
form<strong>at</strong>ives causes the clash <strong>of</strong> the fe<strong>at</strong>ures in th<strong>at</strong> c<strong>at</strong>egory. In<br />
addition, neg<strong>at</strong>ive polarity is dominantly realised by a derived<br />
neg<strong>at</strong>ive lexeme, so th<strong>at</strong> it no longer particip<strong>at</strong>es in the verb’s<br />
primary inflectional paradigm Otoguro , .<br />
Lexical Functional analysis<br />
The preceding sections have revealed th<strong>at</strong> modality is expressed in<br />
diverse ways both in terms <strong>of</strong> their formal encoding, i.e. auxiliaries vs<br />
verbal affixes, <strong>and</strong> their contribution to the syntactic structure, i.e.<br />
fe<strong>at</strong>ures vs predic<strong>at</strong>es. To account for the behaviours modal<br />
expressions show, I will present an analysis in the gramm<strong>at</strong>ical<br />
framework <strong>of</strong> Lexical Functional Grammar LFG, particularly<br />
focusing on the contrast between genuine modals <strong>and</strong> quasimodals in<br />
Japanese.<br />
<strong>Morphosyntax</strong> <strong>of</strong> <strong>Modals</strong> <strong>and</strong> <strong>Quasimodals</strong><br />
LFG is a constraintbased unific<strong>at</strong>ion grammar in which different<br />
linguistic structures are postul<strong>at</strong>ed in a parallel fashion. The surface<br />
organis<strong>at</strong>ions <strong>of</strong> lexical items are represented in constituentstructure<br />
cstructure where words <strong>and</strong> phrases are grouped in a phrase<br />
structure tree <strong>and</strong> encode linear precedence <strong>and</strong> hierarchical
dominance rel<strong>at</strong>ionships. Alongside cstructure, gramm<strong>at</strong>ical rel<strong>at</strong>ions<br />
GF :gramm<strong>at</strong>ical function in LFG <strong>and</strong> other syntactic fe<strong>at</strong>ures are<br />
described in another structure called functionalstructure f<br />
structure. Fstructure is represented as an <strong>at</strong>tributevalue m<strong>at</strong>rix<br />
AVM, inwhich a pair <strong>of</strong> <strong>at</strong>tribute <strong>and</strong> value for a given fe<strong>at</strong>ure is<br />
listed. For instance, a sentence, the children may e<strong>at</strong> breakfast, is<br />
represented in cstructure as in a <strong>and</strong> the corresponding f<br />
structure is given as in b.<br />
<br />
<strong>Morphosyntax</strong> <strong>of</strong> <strong>Modals</strong> <strong>and</strong> <strong>Quasimodals</strong> +,0<br />
The <strong>at</strong>tribute <strong>and</strong> value pairs represented in fstructure come from<br />
the lexical items loc<strong>at</strong>ed in the terminal nodes in cstructure. For<br />
example, is sample lexical entries <strong>of</strong> the items. Each entry<br />
consists <strong>of</strong> a bundle <strong>of</strong> fe<strong>at</strong>ures called functional description as well as<br />
a phrase structure c<strong>at</strong>egories such as Infl, V, D <strong>and</strong> N. is an<br />
abbrevi<strong>at</strong>ion <strong>of</strong> function fM in which f maps one cstructure<br />
node to fstructure, M maps one cstructure node to its mother node,<br />
<strong>and</strong> refers to the ‘current node’ in cstructure as defined in .<br />
Th<strong>at</strong> is, TENSE FUTURE means th<strong>at</strong> in the fstructure<br />
corresponding to the mother node <strong>of</strong> will, i.e. an I node, the value <strong>of</strong>
+,/<br />
TENSE <strong>at</strong>tribute is FUTURE.<br />
a. will I TENSE FUTURE<br />
b. the D DEF <br />
c. children N PRED ‘children’<br />
d. e<strong>at</strong> V PRED ‘e<strong>at</strong>SUBJ,OBJ’<br />
: f M <br />
:f <br />
The m<strong>at</strong>hem<strong>at</strong>ical functions, <strong>and</strong> , are also annot<strong>at</strong>ed on the c<br />
structure, so th<strong>at</strong> the <strong>at</strong>tribute <strong>and</strong> value pairs associ<strong>at</strong>ed with lexical<br />
items are correctly mapped onto fstructure. The annot<strong>at</strong>ed c<br />
structure <strong>and</strong> the corresponding fstructure are illustr<strong>at</strong>ed as in .<br />
<br />
<strong>Morphosyntax</strong> <strong>of</strong> <strong>Modals</strong> <strong>and</strong> <strong>Quasimodals</strong><br />
One <strong>of</strong> the crucial aspects <strong>of</strong> the analysis in is the tre<strong>at</strong>ment <strong>of</strong><br />
the tense auxiliary, will. As specified in the annot<strong>at</strong>ions on the I node
<strong>and</strong> the VP node, i.e. ,the auxiliary <strong>and</strong> the VP is mapped onto<br />
the same outermost fstructure. As a result, the auxiliary merely<br />
contribute the TENSE fe<strong>at</strong>ure to the corresponding fstructure while it<br />
takes a VP complement in the cstructure. Hence, the separ<strong>at</strong>ion <strong>of</strong><br />
surface phrase structure <strong>and</strong> gramm<strong>at</strong>ical fe<strong>at</strong>ure represent<strong>at</strong>ion, LFG<br />
allows us to st<strong>at</strong>e the fact th<strong>at</strong> the auxiliary heads the phrase<br />
structure constituent by taking a VP complement while functionally<br />
it simply adds tense inform<strong>at</strong>ion to the sentence.<br />
Following the observ<strong>at</strong>ion made in section , Falk proposes<br />
the LFG analysis <strong>of</strong> the progressive auxiliary along the line <strong>of</strong> <br />
<strong>and</strong> .<br />
<strong>Morphosyntax</strong> <strong>of</strong> <strong>Modals</strong> <strong>and</strong> <strong>Quasimodals</strong> +,.<br />
a. were I PRED ‘beXCOMPSUBJ’<br />
TENSE PAST<br />
SUBJ XCOMP SUBJ<br />
b. e<strong>at</strong>ing V PRED ‘e<strong>at</strong>SUBJ,OBJ’<br />
ASP PROG
+,-<br />
<br />
<strong>Morphosyntax</strong> <strong>of</strong> <strong>Modals</strong> <strong>and</strong> <strong>Quasimodals</strong><br />
Unlike the future tense auxiliary will, were has its own PRED fe<strong>at</strong>ure,<br />
which suggests th<strong>at</strong> it is functionally an argumenttaking predic<strong>at</strong>e.<br />
Furthermore, one <strong>of</strong> its governable GFs is an open complement called<br />
XCOMP. An XCOMP lacks one <strong>of</strong> its GFs required by the local predic<strong>at</strong>e,<br />
e<strong>at</strong>ing in the case <strong>of</strong> . Such a missing GF is identified with one <strong>of</strong><br />
the GFs in the outer fstructure. This is <strong>at</strong>tained by the mechanism<br />
called functional equ<strong>at</strong>ion, SUBJ XCOMP SUBJ, inthe lexical<br />
entry <strong>of</strong> were. This equ<strong>at</strong>ion ensures th<strong>at</strong> the value <strong>of</strong> SUBJ in the f<br />
structure corresponding to the I is the same as the value <strong>of</strong> the<br />
XCOMP’s SUBJ. This identific<strong>at</strong>ion <strong>of</strong> the values <strong>of</strong> two GFs is represented<br />
by the line connecting them in fstructure.<br />
Another notable fe<strong>at</strong>ure <strong>of</strong> this analysis is a nonthem<strong>at</strong>ic SUBJ
argument <strong>of</strong> the PRED value <strong>of</strong> were in a. The reason th<strong>at</strong> SUBJ is<br />
outside the angled brackets is th<strong>at</strong> it is semantically vacuous to the<br />
predic<strong>at</strong>e. Hence, an XCOMP is the sole semantic argument <strong>and</strong> the<br />
main role <strong>of</strong> the SUBJ is to function as a them<strong>at</strong>ic argument <strong>of</strong> the<br />
inner lexical verb.<br />
The essence <strong>of</strong> this analysis is th<strong>at</strong> the cstructure is minimally<br />
different from , so th<strong>at</strong> it also captures the fact th<strong>at</strong> the<br />
progressive auxiliary were takes a VP complement in the same way as<br />
will. Still, the corresponding fstructure significantly differs from the<br />
AUX fe<strong>at</strong>ure analysis.<br />
I will now present an analysis <strong>of</strong> Japanese modals <strong>and</strong> quasi<br />
modals. An example <strong>of</strong> genuine modals is given as in <strong>and</strong> a<br />
lexical entry <strong>of</strong> the modal expression is specified as in . As<br />
illustr<strong>at</strong>ed in section , the genuine modals in Japanese are inflectional<br />
suffixes, so the entries are given based on the fully inflected forms,<br />
th<strong>at</strong> is the suffixes contribute the value <strong>of</strong> MOOD to the verbs’ lexical<br />
entries.<br />
<strong>Morphosyntax</strong> <strong>of</strong> <strong>Modals</strong> <strong>and</strong> <strong>Quasimodals</strong> +,,<br />
w<strong>at</strong>asiga kono hono yakusoo.<br />
INOM this bookACC transl<strong>at</strong>eINTENT<br />
‘I am willing to transl<strong>at</strong>e this book.’<br />
yakusoo V PRED ‘yakusSUBJ,OBJ’<br />
MOOD INTENT<br />
The cstructure <strong>and</strong> the corresponding fstructure can be represented<br />
as in . Since the genuine modal, oo, does not have an
+,+<br />
independent cstructure node, its fe<strong>at</strong>ure is mapped onto fstructure<br />
together with the lexical verb’s PRED.<br />
<br />
<strong>Morphosyntax</strong> <strong>of</strong> <strong>Modals</strong> <strong>and</strong> <strong>Quasimodals</strong><br />
We now look <strong>at</strong> quasimodals. Since the quasimodals are<br />
independent lexemes inflecting for tense <strong>and</strong> polarity <strong>and</strong> taking a<br />
complement, they have a PRED fe<strong>at</strong>ure subc<strong>at</strong>egorises for a them<strong>at</strong>ic<br />
open complement, XCOMP, <strong>and</strong> a nonthem<strong>at</strong>ic SUBJ. Similarly to the<br />
raising verbs, it requires an equ<strong>at</strong>ion SUBJ XCOMP SUBJ,<br />
which identifies the nonthem<strong>at</strong>ic SUBJ with the SUBJ <strong>of</strong> the inner<br />
PRED. Therefore, the lexical entries for the two quasimodals<br />
appearing in can be given as in .<br />
sytyooga susio taberu tumorid<strong>at</strong>ta rasii.<br />
CEONOM sushiACC e<strong>at</strong> INTENET.PAST REPORT<br />
‘I heard th<strong>at</strong> the CEO was willing to e<strong>at</strong> sushi.’<br />
a. rasii Aux PRED ‘rasiXCOMPSUBJ’<br />
TENSE NONPAST<br />
SUBJ XCOMP SUBJ
. tumorid<strong>at</strong>ta Aux PRED ‘tumoridXCOMPSUBJ’<br />
TENSE PAST<br />
SUBJ XCOMP SUBJ<br />
Those lexical entries allow us to construct the cstructure <strong>and</strong> the<br />
corresponding fstructure for the sentence as in .<br />
<br />
<strong>Morphosyntax</strong> <strong>of</strong> <strong>Modals</strong> <strong>and</strong> <strong>Quasimodals</strong> +,*<br />
Iassume th<strong>at</strong> quasimodals are adjoined to VP in cstructure, <strong>and</strong> the<br />
complement VP is mapped onto the value <strong>of</strong> XCOMP in the fstructure.
++3<br />
The resultant fstructure contains multiple layers <strong>of</strong> XCOMPs, whose<br />
nonthem<strong>at</strong>ic SUBJs are identified with the innermost SUBJ. Note th<strong>at</strong><br />
each layer has a distinct TENSE fe<strong>at</strong>ure, since the quasimodals<br />
independently inflect for tense fe<strong>at</strong>ure.<br />
One intriguing consequence the present analysis predicts is the<br />
relevance between an XCOMP <strong>and</strong> the demonstr<strong>at</strong>ive, sono. As pointed<br />
out by Inoue , averbal pr<strong>of</strong>orm soo is used in a reply referring<br />
to the quasimodal in a question. For instance, in the following<br />
convers<strong>at</strong>ion, soodesu refers to the quasimodal, tumoridesu as shown<br />
in . Toconfirm this observ<strong>at</strong>ion, if we use soodesu as a reply to<br />
the question without a quasimodal, it will be unacceptable as in <br />
B . Tomake the reply acceptable, the speaker needs to repe<strong>at</strong> the<br />
main verb as in B . Even when a genuine modal is used in the<br />
question as in A, soodesu cannot be used as a reply.<br />
A: kimiwa asu daigakuni kuru<br />
youTOPIC tomorrow universityto come<br />
tumoridesuka?<br />
INTENT.POLQ<br />
‘Are you thinking about coming to the university<br />
tomorrow?’<br />
B: hai, soodesu.<br />
yes<br />
<strong>Morphosyntax</strong> <strong>of</strong> <strong>Modals</strong> <strong>and</strong> <strong>Quasimodals</strong><br />
‘Yes, I am thinking about doing so.’<br />
A: kimiwa kesano simbuno<br />
youTOPIC this morningGEN newspaperACC
yomimasitaka?<br />
read.POL.PASTQ<br />
‘Did you read this morning’s newspaper?’<br />
B :??hai, soodesu.<br />
Yes<br />
B :hai, yomimasita.<br />
yes, read.POL.PAST<br />
‘Yes. I read it.’<br />
A: Taroowa ikudesyooka?<br />
TaroTOPIC goCONJECT.POLQ<br />
‘Is Taro going there ?’<br />
B :??hai, soodesu.<br />
yes<br />
B :hai, ikudesyoo.<br />
yes, goCONJECT.POL<br />
‘Yes, he is going.’<br />
This empirical fact indic<strong>at</strong>es th<strong>at</strong> a verbal pr<strong>of</strong>orm soo corresponds to<br />
the fstructure whose PRED is originally a quasimodal. In addition,<br />
the fact th<strong>at</strong> soodesu cannot refer to the modality expressed by a<br />
genuine modal supports the claim th<strong>at</strong> genuine modals in Japanese<br />
are not independent lexical items.<br />
<strong>Morphosyntax</strong> <strong>of</strong> <strong>Modals</strong> <strong>and</strong> <strong>Quasimodals</strong> ++2<br />
A further intriguing aspect is found in the use <strong>of</strong> a genitive<br />
pronoun sono. Acomplement originally introduced by a lexical verb<br />
can be pronominalised by sono <strong>and</strong> used as a nominal complement <strong>of</strong><br />
a quasimodal. This point can be illustr<strong>at</strong>ed by the following
++1<br />
examples.<br />
A: kimiwa asu daigakuni kuru<br />
youTOPIC tomorrow universityto come<br />
tumoridesuka?<br />
INTENT.POLQ<br />
‘Are you thinking about coming to the university<br />
tomorrow?’<br />
<strong>Morphosyntax</strong> <strong>of</strong> <strong>Modals</strong> <strong>and</strong> <strong>Quasimodals</strong><br />
B: hai, sono tumoridesu<br />
yes it INTENT.POL<br />
‘Yes, I am thinking about doing so.’<br />
A: Taroowa kesa tyoosyokuo tabemasitaka?<br />
TaroTOPIC this morning breakfastACC e<strong>at</strong>.POLQ<br />
‘Did Taro have breakfast this morning?’<br />
B: hai, sono hazudesu.<br />
yes it CONJECT.POL<br />
‘Yes, he must have had th<strong>at</strong>.’<br />
In B, asu daigakuni kuru ‘coming to the university tomorrow’ is<br />
pronominalsed by sono <strong>and</strong> becomes a complement <strong>of</strong> the quasi<br />
modal, tumoridesu. Even when a quasimodal is not introduced in the<br />
question, it can be introduced in the reply by taking a pronominal<br />
sono as a complement as shown in B where the quasimodal<br />
hazudesu takes sono, which is a pronominal <strong>of</strong> kesa tyoosyokuo tabeta<br />
‘e<strong>at</strong>ing breakfast this morning’, as a complement. The d<strong>at</strong>a clearly<br />
shows th<strong>at</strong> the quasimodals in Japanese are complementtaking
predic<strong>at</strong>es, <strong>and</strong> in our analysis, soo corresponds to the XCOMP<br />
introduced by a lexical verb, <strong>and</strong> a quasimodal takes it as a<br />
complement.<br />
Conclusion<br />
On the surface, modality is morphosyntactically encoded in various<br />
ways across languages. However, the conceptual consider<strong>at</strong>ion <strong>and</strong><br />
empirical observ<strong>at</strong>ion presented in this paper have revealed th<strong>at</strong> the<br />
modal expressions can be divided into two types, as reflected in the<br />
gener<strong>at</strong>ive liter<strong>at</strong>ure. Since LFG postul<strong>at</strong>es parallel syntactic<br />
structures, it enables us to capture different morphosyntactic<br />
behaviours exhibited by different types <strong>of</strong> modal expressions <strong>at</strong> two<br />
distinct syntactic levels, cstructure <strong>and</strong> fstructure. As argued in<br />
Falk , English has cstructurally similar, but fstructurally<br />
different, auxiliaries. I have also illustr<strong>at</strong>ed th<strong>at</strong> the two distinct types<br />
<strong>of</strong> modal expressions in Japanese require two different analyses in<br />
LFG. The analyses correctly predict the behaviours observed in the<br />
use <strong>of</strong> a verbal pr<strong>of</strong>orm <strong>and</strong> a pronominal in rel<strong>at</strong>ion to the quasi<br />
modals. The present study suggests th<strong>at</strong> the microvariants <strong>of</strong> phrase<br />
structures <strong>and</strong> functions <strong>of</strong> modal expressions found in a diverse<br />
range <strong>of</strong> languages may be captured in a parallel architecture posited<br />
in LFG, which potentially sheds light on typological issues <strong>of</strong><br />
modality.<br />
<strong>Morphosyntax</strong> <strong>of</strong> <strong>Modals</strong> <strong>and</strong> <strong>Quasimodals</strong> ++0<br />
Notes<br />
I would like to thank Yasunari Harada, Tomoko Ohkuma, Sachiko<br />
Shudo <strong>and</strong> Hiroshi Umemoto for valuable discussion <strong>of</strong> the ideas<br />
presented in this paper. This paper is a part <strong>of</strong> the outcome <strong>of</strong> research
++/<br />
<strong>Morphosyntax</strong> <strong>of</strong> <strong>Modals</strong> <strong>and</strong> <strong>Quasimodals</strong><br />
performed under a <strong>Waseda</strong> University Grant for Special Research Projects<br />
Project number : B.<br />
The following abbrevi<strong>at</strong>ions are used throughout this paper : / /,<br />
first/second/third person ; ADV, advers<strong>at</strong>ive ; AUX, auxiliary ; COND,<br />
conditional ; CONJECT, conjective ; COP, copula ; DAT, d<strong>at</strong>ive ; DS, different<br />
subject ; FUT, future ; GEN, genitive ; IMP, imper<strong>at</strong>ive ; INDIC, indic<strong>at</strong>ive ;<br />
INTENT, intentional ; EXCL, exclam<strong>at</strong>ive, EXPL, exempl<strong>at</strong>ive ; IRR, irrealis ; NEG,<br />
neg<strong>at</strong>ive ; NPST, nonpast ; OBLIG, oblig<strong>at</strong>ive ; PASS, passive ; PL, plural ; POL,<br />
polite ; PRES, present ; Q, question ; REPORT, report<strong>at</strong>ive ; SG, singular ; SIM,<br />
simultaneous ; SUBJUNCT, subjunctive.<br />
Quasimodal is not a special term in Japanese linguistics. Chapin ,<br />
for example, uses this term to refer to a set <strong>of</strong> periphrastic modal<br />
expressions in English see also Collins .<br />
The cstructure configur<strong>at</strong>ion <strong>of</strong> Japanese is r<strong>at</strong>her controversial.<br />
Although most <strong>of</strong> the works on Japanese in LFG liter<strong>at</strong>ure generally agree<br />
th<strong>at</strong> Japanese lacks functional projections like IP, whether it has a VP or<br />
not is an open question. In this paper, I adopt the VP structure, but<br />
assuming the nonVP structure would not affect the argument.<br />
References<br />
Bresnan, J. . Lexicalfunctional syntax. Oxford : Blackwell.<br />
Bybee, J., Perkins, R. & Pagliuca, W. . The evolution <strong>of</strong> grammar : Tense,<br />
aspect, <strong>and</strong> modality in the languages <strong>of</strong> the world. Chicago : The University <strong>of</strong><br />
Chicago Press.<br />
Chapin, P.G. . <strong>Quasimodals</strong>. Journal <strong>of</strong> Linguistics . .<br />
Chomsky, N. . Aspects <strong>of</strong> the theory <strong>of</strong> syntax. Cambridge, MA : MIT Press.<br />
Cinque, G. . Adverbs <strong>and</strong> functional heads : A crosslinguistic perspective.<br />
Oxford : Oxford University Press.<br />
Collins, P. . <strong>Modals</strong> <strong>and</strong> quasimodals in English. Amsterdam : Rodopi.<br />
Dyvik, H. . The universality <strong>of</strong> fstructure : Discovery or stipul<strong>at</strong>ion?<br />
the case <strong>of</strong> modals. In Butt, M. & King, T.H. eds., Proceedings <strong>of</strong> the LFG<br />
+333 conference, Stanford, CA : CSLI Public<strong>at</strong>ions.<br />
Falk, Y. . Functional rel<strong>at</strong>ions in the English auxiliary system.<br />
Linguistics . .<br />
Harbert, W. . The Germanic languages. Cambridge : Cambridge
<strong>Morphosyntax</strong> <strong>of</strong> <strong>Modals</strong> <strong>and</strong> <strong>Quasimodals</strong> ++.<br />
University Press.<br />
Inoue, K. . Nihongo no modal saiko Japanese modals revisited. In<br />
Hasegawa, N. ed., Nihongo no shubun genshoo Clausal phenomena in<br />
Japanese, Tokyo : Hituzi Shoboo. .<br />
Jackend<strong>of</strong>f, R. . Toward an explan<strong>at</strong>ory semantic represent<strong>at</strong>ion.<br />
Linguistic Inquiry . .<br />
Jackend<strong>of</strong>f, R. . X…syntax : A study <strong>of</strong> phrase structure. Cambridge, MA :<br />
MIT Press.<br />
Kiefer, F. . Ondefining modality. Folia Linguistica . .<br />
Kroeger, P.R. . Analyzing grammar : An introduction. Cambridge :<br />
Cambridge University Press.<br />
Lyons, J. . Introduction to theoretical linguistics. Cambridge : Cambridge<br />
University Press.<br />
Lyons, J. . Semantics. Cambridge : Cambridge University Press.<br />
Mithun, M. . The languages <strong>of</strong> N<strong>at</strong>ive North America. Cambridge :<br />
Cambridge University Press.<br />
Narrog, H. . Ondefining modality again. Language Sciences . .<br />
Nitta, Y. . Nihongo no modality to ninshoo Japanese modality <strong>and</strong> person.<br />
Tokyo : Hituzi Shoboo.<br />
Otoguro, R. . Paradigm gaps <strong>and</strong> periphrases in the Japanese<br />
conjug<strong>at</strong>ion system. In Proceedings <strong>of</strong> western conference on linguistics ,**1.<br />
University <strong>of</strong> California, San Diego. .<br />
Otoguro, R. . Surfaceoriented morphological network. Humanitas .<br />
. The <strong>Waseda</strong> University Law Associ<strong>at</strong>ion.<br />
Palmer, F.R. . Mood <strong>and</strong> modality. Cambridge : Cambridge University<br />
Press, second edition.<br />
Pollock, Y. . Verb movement, Universal Grammar, <strong>and</strong> the structure <strong>of</strong><br />
IP. Linguistic Inquiry . .<br />
Ross, J.R. . Auxiliaries are main verbs. In Todd, W. ed., Studies in<br />
philosophical linguistics, Evanston, IL : Gre<strong>at</strong> Expect<strong>at</strong>ions Press. .<br />
Schwarze, C. . The syntax <strong>of</strong> Romance auxiliaries. In Butt, M. & King, T.<br />
H. eds., Proceedings <strong>of</strong> the LFG30 conference, Stanford, CA : CSLI<br />
Public<strong>at</strong>ions.<br />
Ueda, Y. . Nihongo no modality no toogo koozoo to ninsyoo hyoogen<br />
Syntax <strong>of</strong> Japanese modality <strong>and</strong> person restrictions. In Hasegawa, N.
<strong>Morphosyntax</strong> <strong>of</strong> <strong>Modals</strong> <strong>and</strong> <strong>Quasimodals</strong> ++-<br />
ed., Nihongo no shubun genshoo Clausal phenomena in Japanese, Tokyo :<br />
Hituzi Shoboo. .<br />
Wright, S. . Subjectivity <strong>and</strong> experiential syntax. In Stein, D. & Wright,<br />
S. eds., Subjectivity <strong>and</strong> subjectivis<strong>at</strong>ion : Linguistic perspectives, Cambridge :<br />
Cambridge University Press. .